Corrosive environments have a significant detrimental impact on aircraft engine turbine blades resulting in early degradation and a higher risk of failure. Currently, human visual inspection evaluates the condition of blades and identify premature degradation such as cracking or corrosion. While this approach works, it is time-consuming to carry out manual examinations and susceptible to human error. More so, it lacks a robust and objective strategy to identify the conditions of the turbine in terms of thermal cycle and exposure. Instead, machine learning approaches have ample potential to identify and quantify degradation from images and classify damage conditions in a robust and economical manner. Hence, this study explores the use of deep neural networks to determine the environment to which a nickel-base superalloy was exposed in laboratory testing. A machine learning approach was implemented to predict temperature, salt flux, material type and exposure times using a database with 3000 images of sample cross sections. We compared two machine learning environments (MATLAB, and Python) and we enriched the database by cropping images. The results demonstrate that machine learning approaches have impressive predictive power for laboratory samples that can sometimes be superior to that of human experts. We further identify the environmental attributes that are more difficult to predict and which predictions can be achieved confidently.